Every Patient Tells a Story (37 page)

The authors of the Google study note that, in fact, Google is likely to be a more precise diagnostic tool for clinicians than the lay public because clinicians will use more specific search terms (“myocardial infarction” rather than “heart attack,” for example) and will be better able to identify likely hits because of their preexisting knowledge. Patients, using everyday language, are likely to end up with fewer useful hits buried in pages of irrelevant sites. Their ability to distinguish the useful hits will be compromised by their unfamiliarity with medical language.

The power of Google in the realm of medical diagnosis has not been lost on Google itself. Google has formed a Health Advisory Panel to inform its work in this area. And Google has launched a major effort to improve the quality of medical-related searches by having reputable organizations (such as the National Library of Medicine) and individual doctors flag Internet sites offering reliable information. These sites are then given prominence when search results are returned and are labeled with the individual or organization that has vetted them.

Google is very open about its plans to improve search capabilities for patients, but the company is mum on the subject of doing the same thing for physicians (Google representatives declined to be interviewed on this subject). Perhaps that’s because doctors are a valuable audience and if Google can find a way to improve diagnostic search results to the point of being more accurate than Isabel and other commercial systems, it could effectively capture the market and be able to leverage all those physician “eyeballs” with advertisers.

But even a more accurate Google-based diagnostic decision support system
wouldn’t really solve the problem of missed diagnoses. To begin with, any system that must be consulted separately from the digital workspace in which a doctor or nurse deals with a patient will only be used when there is uncertainty in the mind of the health care professional. If a doctor is sure of her diagnosis, or a nurse is certain that the correct medication has been prescribed, they won’t turn to Google (or Isabel, or DXplain, or any other system).

Computer programs won’t really make a dent in the problem of misdiagnoses and other types of medical errors until they are much “smarter” and easier to use than they are today.

“Future systems need to operate in the background,” says Eta Berner, the researcher who has tracked progress in medical computing for decades. “The doctor shouldn’t have to enter anything. The system should be able to extract information from what the doctor or nurse is already doing … taking notes or entering lab values or prescribing medications. The system should be intelligent enough to provide an alert or a reminder only if something is really missing … a test, for example, or a medication.”

Berner foresees a time when all of the now fragmented information streams in the health system will be unified and made consistent. Patients’ health records will be fully digital—including images such as MRI scans or X-rays. Standard words, phrases, and units of measurement or description will be used so that computer systems in distant locations can intelligently and accurately use the information. Doctors and nurses will enter all information in digital form—handwriting (never doctors’ strong suit anyway) will be obsolete.

With this kind of a system in place, the possibility of infection with the schistosomiasis parasite would have popped up the very first time the young woman described earlier was evaluated in an emergency room. The likelihood that little Isabel Maude was suffering from a rare complication of chickenpox would not have been easy to ignore. And the patient with Rocky Mountain spotted fever wouldn’t have had to use Google herself … her doctor would have already seen the tight fit between her symptoms and that possible diagnosis.

Of course it will be years—and more likely decades—before this kind
of a system is in place. And although I think it is inevitable that the vast resources of the digital age will become more fully integrated into our health care system and the doctor’s diagnostic routine, it may not take the form we anticipate. Computers have already revolutionized our diagnostic abilities dramatically. I think the first and most important digital diagnostic tool developed was the CT scanner. It was the development of powerful computers that allowed us to capture data from a series of two-dimensional images to create a three-dimensional representation of the body. Since 1972, when the CT scan was first developed, this tool has made routine diagnoses that would previously only have been discovered after death. So while we envision a future where the computer learns how to think like a doctor, it is possible that its greatest contributions will take a very different form.

Would a kind of super-efficient, integrated, intelligent computer system eliminate all diagnostic challenges? Would it replace doctors? Hardly. I believe the process of diagnosis will be made more effective and that it will be faster and easier in the future to zero in on what’s really wrong with a patient. But there will always be choices to make—between possible diagnoses, between tests to order, and between treatment options. Only a skilled and knowledgeable human can make those kinds of decisions.

And, of course, people need more than the right treatment for the right disorder. They need to be heard, they need reassurance, explanations, encouragement, sympathy—the full range of emotional support that is a critical part of what we doctors try to do: heal.

AFTERWORD
The Final Diagnosis

“I
’m sorry,” the young man on the telephone said to me. His voice was hushed and sympathetic, difficult to hear over the usual commotion of the clinic bustling just outside my office door. He was a stranger to me. He said his name was Jorge. He was an old friend of a young woman we both knew quite well. “I’d chatted with her on the phone maybe twenty minutes earlier. She said come by and so I just drove on over.”

He told me that he’d rung her bell early that sunny September morning and when there was no answer, he clanged through the backyard gate. When he saw her stretched out on the chaise longue in her bathing suit, his first thought was how pretty she looked. “I’m a married man, so it wasn’t like that, but she’s always been a looker.” When she didn’t reply to his “Hey, how’s it going?” he approached her and put his hand on her shoulder. Her skin felt warm but he noticed how strangely pale she was under her tan. “And I knew then, I knew. Her cell phone was right there next to her, like it always was, so I picked it up and dialed 911.”

I thought back to the last time I’d seen Julie: her tanned cheeks still un-lined, her eyes so blue that even the whites were the color of robin’s eggs. I could hear her deep tobacco-coarsened drawl and her earthy sense of humor. I closed my office door and dropped into my chair.

My beautiful and mysterious little sister was dead.

My first thought, when thought was finally possible, was how? More than anything, I wanted to know how a young woman could die so suddenly that she didn’t even have time to call for help. What happened?

It was a strangely familiar question. When patients of mine have died, their spouse or parent or child or friend would ask me this very question after I broke the news. In waiting rooms outside the emergency room or ICU, shocked, sad, crying—they would ask: Doctor, how did this happen? How did this person, so very alive not so long ago, die? I would do my best to answer, to pull together the strands of a devastating illness or collapse, but it seemed a peculiar question—as if an explanation could somehow soothe the jagged edges of loss. But it made sense to me now. I suddenly understood that terrible need to know how.

At forty-two, my sister was healthy. But she was also an alcoholic. For the past fifteen years or so, her life had been dominated by this desire, and then this need, to drink. She’d started out—like so many—with excesses in high school, but calmed down after marriage and the birth of the son she loved. Over time, and for reasons I will never know, Julie’s drinking became more frequent. Weekend binges rapidly became the daily dose she’d sneak as she got her son ready for day care, or as she set out for work, as she prepared dinner or put her son to bed.

She tried to stop. Again and again she would check herself into a hospital, or simply start going to AA meetings and try—I think, really try—to stop. She would call us almost daily, triumphant with the exact number of days, even hours, since her last drink. Then the calls would become less frequent. Her voice mail would tell us she’d call us back but she rarely did. And then finally there would be silence. Until she would try once more. My sisters and I—we were a family of five sisters—watched in helpless distress. Over the years we’d learned what all relatives of alcoholics learn: that everything we could do still was not enough.

And then she died, as mysteriously as she had lived.

What could kill a young woman that young, that fast? Jorge had found
her cell phone along with a pack of cigarettes and a Coke sitting right beside her. She was obviously tanning herself, relaxing in the summer sun. Whatever killed her struck so quickly that she could not reach over to pick up the phone and dial 911. What could do
that
? I couldn’t get that terrible question out of my mind. As I made arrangements to travel home, I puzzled over it. I went into my doctor mode—in part because it was a way of managing my grief and in part because it’s what I’m trained to do. And without really wanting to, I found myself putting together a differential diagnosis, searching for scenarios that might explain how my sister had died so abruptly.

Certainly a heart attack can be quick and deadly, especially at a young age. But that would be unusual in a forty-two-year-old woman. And we had no family history of heart disease. A ruptured blood vessel in her brain could cause an instantaneous loss of consciousness and rapid death. A massive clot that went to her lungs was another possibility. She was a smoker; maybe she was also taking birth control pills. That combination has been linked to such clots. Infection seemed unlikely. And yet, had she been sick? I didn’t know. Suicide was unthinkable to me, but it had to remain a possibility. She was often deeply depressed during these relapses. An accidental overdose was also possible.

The coroner in Savannah, Georgia—where she had lived her last year and where she had died—ordered an autopsy be performed. Although one of my sisters was upset with what she saw as a violation, I was grateful. An autopsy, I hoped, would provide me with this necessary and final diagnosis.

Autopsy—the word comes from the Greek
autopsia
, meaning to see for oneself. Historically the autopsy has played a critical role in medicine. For centuries everything we knew about disease was derived from examining the body after death. Even now when my patients ask me about their aches and pains for which I have no diagnosis, I confess to them that our knowledge of diseases that
can’t
kill you is fairly new and much less developed because even now most of what we know about disease was derived postmortem. Medicine’s first toehold into modern-day diagnosis came at the last half
of the eighteenth century, when Giovanni Battista Morgagni, a physician and professor at the University of Padua, published
On the Seats and Causes of Diseases Investigated by Anatomy
. This book, completed when Morgagni was seventy-nine years old, was composed of hundreds of beautifully detailed drawings from autopsies that he’d performed over the course of a long career. These carefully drawn images revealed the destruction and distortions of the anatomy hidden beneath the skin and leading to death. By showing exactly how disease manifests itself in these visible, concrete ways within the body, the work inspired generations of doctors to investigate the process by which disease can distort and derange our most fundamental anatomy. For centuries disease and death had been attributed to humors or spirits or other intangibles and not something as real, or as clearly visible, as it was in these images.

For the past 250 years autopsy has been one of medicine’s most reliable sources of information about the nature of disease. Cancer, heart disease, hemorrhage were all first seen through the exploration of the body after death. In the twentieth century, autopsy was used as the ultimate diagnostic tool. At its peak, up to half of all patients who died in the hospital underwent postmortem evaluation. Too late to help the patient, what was revealed was often useful knowledge for the doctor, the hospital, the family. Diseases missed or undetectable with the available technology were finally made visible. Doctors could use the knowledge for the benefit of their next patients. Hospitals used the information as a form of quality assurance on the care they provided and the skills of the doctors who practiced there. There were benefits for the bereaved family as well. The disease that took their loved one could be a risk for them as well.

These days, patients who die in a hospital rarely make it to the pathologist’s table. Hospitals used to be required to perform autopsies. The Joint Commission on Accreditation of Healthcare Organizations—the regulatory body overseeing hospitals—required these institutions to maintain autopsy rates of at least 20 percent (25 percent for teaching hospitals), which was, and continues to be, the rate most advocates say is the minimum for monitoring diagnostic and hospital error. The commission eliminated that requirement
in 1970. Medicare stopped paying for those that still got done a few years later.

Until quite recently autopsies were also considered an essential component in medical training. Residency programs were required to get autopsies on 15 percent of all the patients who died while under resident care. Seeing the real ravages of disease was considered an important part of medical training. But the requirements for most medical trainees were rolled back in the 1990s. Small residency programs objected to the ever growing cost—autopsies were not paid for—and enforcement of the rule was difficult.

Even before the rollbacks of the requirements on hospital and training programs, the number of autopsies performed had plummeted. In the 1960s, nearly half of those who died in the hospital were autopsied. Only forty years later, at the turn of the twenty-first century, that rate had dropped to less than six per one hundred in-hospital deaths. We don’t even know how many are done now because that data isn’t collected anymore. In the community hospital where I take care of patients, there were ninety-three autopsies done in 1983. One recent year, we had performed a grand total of eleven autopsies and almost half of those were on stillborn infants.